The Empty Array: In Indian Chess, the Audience Grew; the Documents Did Not

At 2:14 a.m. the extraction script finished. Eleven of twelve fields came back empty — title, source, article type, one-line summary, author stance, the list of information points, all of it. One field survived: domain_label: chess. No exception was thrown. No error was logged. The system returned zero rows in a clean, well-formatted way and called it a job done. In data work this is the most expensive kind of failure. A script that crashes gets fixed. A script that returns an empty array and says 'all clear' stays undetected for weeks. I read the logs three times, opened the raw JSON, counted the rows. Same result: nothing. And the nothing came back for a sport that has handed this country four major titles in a decade, whose Olympiad double gold is barely two years old, and whose largest media operation reaches 2.9 million subscribers. My stake here is personal, so let me declare it early. In November 2026, in a Kolkata press box, I asked for the first time for the PGN and clock times of the Tata Steel Chess India rapid games. What came back was a tilted photograph of a handwritten scoresheet on WhatsApp. There were two women in a room of forty men, and a senior reporter told me chess tactics were 'not a women's beat'. I answered with a spreadsheet — centipawn loss and clock usage across all 45 rapid games, Anand included. Since that day my rule has been simple: I do not write a sentence that cannot carry a footnote. Eight years later the same question returned as a script, and the answer was empty again. A chess record is built from five things, and none of them is optional: the event identity and date; both players' names and their ratings at the time; the move-by-move PGN; clock time per move; and the arbiter's verification. A sixth element attaches later — the organisational record: who selected the team, on what criteria, who paid, who coached. Compared with cricket or football, chess has one advantage: its data is machine-readable. A single PGN file holds an entire game; add clock times and you get a map of the decisions taken under pressure. The cost is that none of this data generates itself. Somebody has to decide to write it down. FIDE published its first official rating list in July 2026; since then the monthly list has been an account nobody can privately own. But the rating list will never tell you what happened in the final round of that Kolkata hall. The specific problem here is that titles are produced faster than files. Look at 2026 to 2026: Anand–Carlsen in Chennai (November 2026, Carlsen won 6.5–3.5), the shared Online Olympiad title (August 2026), the Chennai Olympiad (28 July to 10 August 2026), India's first double gold in the open and women's sections at Budapest (September 2026), and Gukesh's world title in Singapore (12 December 2026). Eleven years, five major events. How many of those have a complete public file? The PGNs survive, on organiser sites and results platforms. Clock times survive for some events, incompletely. Selection criteria, coaching staff, the money ledger — almost nothing. I keep a dated list of AICF elections, state-association suspensions and state cash awards going back to 2026, updated every quarter. That list lives on my laptop. There is no public version of it, and until another reporter builds one, every new dispute will be written without comparison to the last. A methodology note, because that is my habit: sample — five events between 2026 and 2026; source — my own PGN-and-clock archive plus public results; limitation — I have no access to organisational or financial documents, so my claim about their absence rests on a search, not on an inference. November 2026, Kolkata. Forty-five rapid games, ten players, one round robin. I wanted three columns: centipawn loss per move, clock time per move, and average time taken in the last five moves before the time control. Three columns in, the story of the event changed — the weak moves were not arriving in the opening but under the clock. That was the insight a photograph of a scoresheet could never deliver. The real finding was not in the spreadsheet, though. It was that the spreadsheet existed only with me. No organiser asked for it, no federation archived it, no publisher turned it into a collection. The spreadsheet was three weeks old before the press box even opened — and it had no home. August 2026, Hyderabad. The FIDE Online Olympiad final dissolved into server failures and India was declared joint winner with Russia. I was working night shifts from my flat. I logged every disconnection against the move number and published an audit of which decisive games had sat on unstable connections. Every contested server leaves a timestamp, if you wait long enough. The same month something else was happening. Streams and comedy-chess shows were multiplying the audience. I distrusted the noise, so I measured it — six straight weeks of tracking concurrent viewers against broadcast numbers. The result was that streaming gained an audience, broadcast lost one, and both numbers were true in the same year. Since then every piece I file carries a methodology note: sample, source, limitation. A viewer count is never a finding; it is a fact about the audience, and it should be used with exactly that much force. My audit stayed in circulation for six weeks, then died. No institution picked it up. Today anyone wanting to argue about the server failures in that 2026 final has my blog and my screenshots, and nothing else. That is the archive crisis in its cleanest form: the fact existed, its home did not. 28 July to 10 August 2026, Chennai. India hosted an Olympiad for the first time. India-2 took open bronze, India-1 took women's bronze, and India took the Gaprindashvili Cup. My editor wanted me in Qatar for the World Cup; I chose Chennai, and the reason was precedent — the 2026 Anand–Carlsen match had been Indian chess journalism's biggest assignment, and 2026 was an event of the same class. In the analysis room an official called the women's section 'the soft bracket'. I did not argue. I pulled board-by-board conversion rates — of the positions that reached a winning evaluation, how many were actually won. The answer was not an argument; it was a table. That is my habit and also my limit: I do not win debates, I only settle them. The same habit told me where the year's weakest record sat. The tournament's PGNs were public. The organisation behind them was not. The expenditure of a publicly funded event, a staffing list, a selection criterion — none of those documents can be found today. Before the match there was a file nobody had opened, and after the match it stayed shut. September 2026, Budapest. India won gold in both the open and women's sections of the same Olympiad for the first time. It is the largest collective result in the country's chess history. Now the colder question: in which document is the criterion for those two teams recorded? Ten players across two squads, board-order decisions, use of reserves — I could not find a public paper for any of it. The result has been written. The process never was. I am not saying the process was compromised. I am saying that had it been, there would be no basis for an enquiry. That is the difference between a document and a memory, and it decides whether a sport can audit itself. In Chennai 2026 Carlsen took the title, and much of that year's journalism rested on reporters' eyes — what was seen was written down. In Singapore 2026 millions watched live and very few wrote it down. Eleven years separate the two events and the direction of the record has reversed: the audience grew, the witnesses shrank. There is arithmetic behind that. When everything is streamed live, the need to write is not felt. The trouble is that a broadcast is not an archive. A VOD without an index is a warehouse, not a library. You cannot pull round four, board seven, move 34 out of it unless somebody has already made it searchable. Journalism builds that index. When journalism thins, footage grows and knowledge shrinks — and the decline happens loudly, which is why nobody hears it. Now the part where I have to argue against myself. The first mistake would be to call everything above an institutional crisis. The truth is that the empty array may well have been my own script's fault — an upstream extraction bug that returned zero instead of a list. Converting a tooling failure into a national failure is exactly the carelessness I criticise. What holds is separate and separately checked: what was not found in a search of public documents was not found. Merging the two conclusions into one sentence would be wrong. The second mistake would be to blame streaming. The numbers do not permit it. 2.9 million subscribers and rising concurrent viewership show that chess in this country has left the tier below cricket in terms of reach. Reach is a fact. It is only a fact, not proof — but it cannot be denied as a fact. The media expansion made chess legible; that is true. It is also true that the expansion did not make a record. The third mistake is to read correlation as cause. Titles and coverage grew in the same decade, so it is easy to say coverage produced the titles. The claim fails testing. Tamil Nadu's pipeline predates 2026 by at least two decades — the chess programme in Velammal schools, the Tal Chess Club lineage, state cash awards, and payrolls at public undertakings like ONGC and Indian Railways. The structure was built from below; the coverage arrived from above. The verified finding is duller and more useful: the media made chess visible, the schools made it possible. The fourth mistake, the one I make most often: assuming every AICF dispute is a repeat of the last one. Ten years of notes create that illusion. So I date-stamp the precedent explicitly — the list of state-association suspensions since 2026 — and then ask what is genuinely new. What is new is that for the first time since 2026, the money and the record of an event sit in different hands. They used to sit at the same table, which at least allowed a reconciliation. Before the next major event, if I get to ask for one thing, it will not be a highlight reel. I will ask for six fields: date, players, PGN, clock times, arbiter verification, and one sentence of selection criteria. Five of those can be produced today at no cost. The sixth is politics, which is why it is hardest. India has world champions; that is on the record. The question is what a researcher of the next generation will find when they open Budapest 2026 — a scoresheet, or an account. I built the dataset first and let the story arrive late; what remains to be decided is who writes the first row.